Ralf Stubner <[EMAIL PROTECTED]> wrote: >> I guess the patch to texconfig that you wrote is >> the better approach. > > The approach is flawed, however,
That's what occurred to me too while riding my bike to the university... > since it is very easy to construct > situations where files in /etc are automatically changed, which we must > not do. Hence it is probably easier to make all files that can be > changed via texconfig proper configfiles in /etc, as you already > suggested. Then we would need the permission to modify these files in > some aspects. We could get this permission either via a debconf message > (IMO overkill) or by something like this: > > The libpaper hock script reads in, say, /etc/default/tetex which > contains a variable, say, USE_LIBPAPER. By default this variable is set > to 'no', which makes the libpaper hock script excit with a suitable > message. If the variable is set to 'yes', the libpaper hock script will > adjust the configuration files. That's a nice approach. No need to hack texconfig even further. We need to - ship the libpaper hook and the file in /etc/default - synchronize the available paper sizes in texconfig and libpaper, filtering unknown ones out in the hook script (even if currently none would be unknown) - ? - test everything Some remarks: * I think we shouldn't use the filename /etc/default/tetex: The setting will actually also have effect when texlive is installed and used * It would be good if texconfig could be notified that it doesn't need to recreate the formats, since we're doing that anyway a couple of lines later. I think such an option would even make sense to submit upstream, because it is usefull if one wants to change a couple of things at the same time, like papersize and hyphenation patterns. * Generally, we should try to provide the parts that make sense upstream as separate patches. > Nice thing would be that no patches for texconfig or dvips/dvipdfm would > be needed. Except enabling more paper sizes - but that's also something upstream will probably accept. Regards, Frank -- Dr. Frank Küster Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich Debian Developer (teTeX/TeXLive)